THE PROJECT SPONSOR
The project sponsor is the intended end user of the application. No consultant is involved.
OVERVIEW OF PROJECT
An application to achieve batch processing of computer image analysis of multiple image files. The image files will exist in a database and new image files will be added periodically in real time. Each image will have a unique ID as well as a file name. The software must monitor for new images as they are added and process them on a FIFO basis. Quantitative measures from the outputs of the analysis should then be posted to the database.
THE IMAGE FILES
A range of standard file types: jpeg, png, bitmap etc. File sizes about 10 kb to 300 KB.
IMAGE ANALYSIS - PARAMETER MEASUREMENT
The appliction will permit a User Defined Analysis. A simple GUI should allow the user to specify which of a range of parameters should be quantified for each image, such as Hue, Saturation, Value, file size, XY pixel dimentions etc.
The user may elect either to process the batch with predefined parameter measurements (a stored default), or to choose which parameters should be quantified for the particular batch.
The User Defined Analysis GUI will allow the user to toggle on/off the option to divide the image into two or three pieces (whole image; top half; botttom third etc) and the application will then calculate the set of chosen parameters for each part of the image and the image as a whole. An emphasis is placed on calculating averages for each parameter for each image component. For example, the average Hue or the average Saturation.
IMAGE ANALYSIS - OBJECT COUNTING (optional)
Developers may choose to present proposals that also provide object counting as an option in the User Defined Analysis. The user may toggle this feature on or off in the analysis. If selected, the application will make a "second pass" analysis through each image with respect to counting discrete objects. Commercial object recognition software might be utilized in this component of the application such as GSA Image Analyzer, but preference will be given to open source solutions. It is understood that errors will be experienced in the object counting... the application simply needs to make good enough counting over a range of sample images... which will be supplied.
This component of the application is a "nice to have" and developers are not require to include it in their proposals. Preference however will be given to developers who can deliver both parameter and object counting.
Possible software for consideration includes, but not limited to: MySQL, ImageMagick, OpenCV, GraphicsMagick, ImageJ, C#